Está en la página 1de 6

Computers and Mathematics with Applications 57 (2009) 757762

Contents lists available at ScienceDirect

Computers and Mathematics with Applications


journal homepage: www.elsevier.com/locate/camwa

B-spline method for solving linear system of second-order boundary


value problems
Nazan Caglar a, , Hikmet Caglar b
a
Istanbul Kultur University Faculty of Economic and Administrative Science, 34156 Atakoy-Istanbul, Turkey
b
Istanbul Kultur University Department of Mathematics-Computer, Istanbul, Turkey

article info a b s t r a c t

Article history: The B-spline method is used for the numerical solution of a linear system of second-order
Received 3 July 2008 boundary value problems. Two examples are considered for the numerical illustration and
Received in revised form 14 July 2008 the method is also compared with the method proposed by J. Lu [J. Lu, Variational iteration
Accepted 10 September 2008
method for solving a nonlinear system of second-order boundary value problems, Comput.
Math. Appl. 54 (2007) 11331138]. The comparison shows that the B-spline method is a
Keywords:
more efficient and effective tool and yields better results.
Second-order boundary value problems
B-spline
2008 Elsevier Ltd. All rights reserved.
Linear system
Differential equation
Boundary conditions

1. Introduction

Systems of ordinary differential equations have been applied to many problems in physics, engineering, biology and so
on. There are many publications dealing with the linear system of second-order boundary value problems. They introduced
various numerical methods. For instance, a finite difference method has been proposed in recent works [26]. Also Geng et al.
have studied the numerical solution of a nonlinear system of second-order boundary value problems in the reproducing
kernel space [7]. J. Lu considered the variational iteration method to solve a nonlinear system of second-order boundary
value problems [1]. In this paper, we present and analyze the B-spline method for the solution of a linear system of second-
order boundary value problems with the assumption that the solutions are unique. We consider the following linear system
of second-order boundary value problems:

u00 + a1 (x)u0 + a2 (x)u + a3 (x)v 00 + a4 (x)v 0 + a5 (x)v = f1 (x),


(
v 00 + b1 (x)v 0 + b2 (x)v + b3 (x)u00 + b4 (x)u0 + b5 (x)u = f2 (x), (1)
u(0) = u(1) = 0, v(0) = v(1) = 0,
where ai (x), bi (x), f1 (x) and f2 (x) are given functions, and ai (x), bi (x) are continuous, i = 1, 2, 3, 4, 5.

2. Analysis of B-spline method

The third-degree B-spline is used to construct numerical solutions to a given problem (1). A detailed description of third-
degree B-spline functions can be found in [8]. The third-degree B-splines are defined as
Bi1 (x) = B0 (x (i 1)h), i = 2, 3, . . . ,

Corresponding author.
E-mail address: ncaglar@iku.edu.tr (N. Caglar).

0898-1221/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.camwa.2008.09.033
758 N. Caglar, H. Caglar / Computers and Mathematics with Applications 57 (2009) 757762

Table 1
Values of Bi , B0i and B00i .

xi xi+1 xi+2 xi+3 xi+4

Bi 0 1 4 1 0
B0i 0 3/h 0/h 3/h 0
B00i 0 6/h2 12/h2 6/h2 0

0x<h
3
x
h x < 2h

1 3x3 + 12hx2 12h2 x + 4h3

B0 (x) = (2)
6h3 3x3 24hx2 + 60h2 x 44h3 2h x < 3h
3h x < 4h.

3
x + 12hx2 48h2 x + 64h3
To solve second-order boundary value problems, Bi , B0i and B00i evaluated at the nodal points are needed. Their coefficients
are summarized in Table 1.
Let
n 1
X
u(x) = Cj Bj (x), (3)
j=3

n1
X
v(x) = Dj Bj (x), (4)
j=3

be an approximate solution of Eq. (1), where Cj and Dj are unknown real coefficients and Bj (x) are third-degree B-spline
functions. Let x0 , x1 , . . . , xn be n + 1 grid points in the interval [a, b], so that
xi = a + ih, i = 0, 1, . . . , n; x0 = a, xn = b, h = (b a)/n. It is required that the approximate solution (3) and (4)
satisfies the differential equation at the points x = xi . Putting Eqs. (3) and (4) in (1), it follows that
n 1
X n 1
X
Cj [B00j (xi ) + a1 (xi )B0j (xi ) + a2 (xi )Bj (xi )] + Dj [a3 (xi )B00j (xi ) + a4 (xi )B0j (xi ) + a5 (xi )Bj (xi )] = f1 (xi )
j=3 j=3

i = 0, 1, . . . , n, (5)
n 1
X n1
X
Dj [B00j (xi ) + b1 (xi )B0j (xi ) + b2 (xi )Bj (xi )] + Cj [b3 (xi )B00j (xi ) + b4 (xi )B0j (xi ) + b5 (xi )Bj (xi )] = f2 (xi )
j=3 j=3

i = 0, 1, . . . , n, (6)
and boundary conditions can be written as
n 1
X
Cj Bj (x) = 0, for x = 0, (7)
j=3

n 1
X
Cj Bj (x) = 0, for x = 1, (8)
j=3

n 1
X
Dj Bj (x) = 0, for x = 0, (9)
j=3

n 1
X
Dj Bj (x) = 0, for x = 1. (10)
j=3

The spline solution of Eq. (1) is obtained by solving the following matrix equation. The value of the spline functions at
the knots {xi }ni=0 are determined using Table 1 and substituting (5)(10). Then a system of 2(n + 3) linear equations in the
2(n + 3) unknowns C3 , C2 , . . . , Cn1 , D3 , D2 , . . . , Dn1 are obtained. This system can be written in the matrix-vector
form as follows
AE = F , (11)
where
E = [C3 , C2 , . . . , Cn1 , D3 , D2 , . . . , Dn1 ]T ,
F = [0, f1 (x0 ), . . . , f1 (xn ), 0, 0, f2 (x0 ), . . . , f2 (xn ), 0]T ,
N. Caglar, H. Caglar / Computers and Mathematics with Applications 57 (2009) 757762 759

and A is an 2(n + 3)x2(n + 3)-dimensional matrix given by


M1 | M2
" #
A= .
M4 | M3
We define four submatrices M1 , M2 , M3 and M4 as follows:
1 4 1 0 0 0

1 (x0 ) 1 (x0 ) 1 (x0 ) 0 0 0
0
1 (x1 ) 1 (x1 ) 1 (x1 ) 0 0

,
M1 =



0 0 0 1 (xn ) 1 (xn ) 1 (xn )

1 4 1
0 0 0 0 0 0

2 (x0 ) 2 (x0 ) 2 (x0 ) 0 0 0
0
2 (x1 ) 2 (x1 ) 2 (x1 ) 0 0

,
M2 =



0 0 0 2 (xn ) 2 (xn ) 2 (xn )

0 0 0
1 4 1 0 0 0

3 (x0 ) 3 (x0 ) 3 (x0 ) 0 0 0
0
3 (x1 ) 3 (x1 ) 3 (x1 ) 0 0


M3 = ,



0 0 0 3 (xn ) 3 (xn ) 3 (xn )

1 4 1
0 0 0 0 ... 0 0

4 (x0 ) 4 (x0 ) 4 (x0 ) 0 0 0
0
4 (x1 ) 4 (x1 ) 4 (x1 ) 0 0

.
M4 =



0 0 0 4 (xn ) 4 (xn ) 4 (xn )

0 0 0
Also the coefficients in the submatrices M1 , M2 , M3 and M4 have the following form
6 3
1 (xi ) = + a1 (xi ) + a2 (xi ), i = 0 , . . . , n,
h2 h
6 3
2 (xi ) = a3 (xi ) + a4 (xi ) + a5 (xi ), i = 0, . . . , n,
h2 h
6 3
3 (xi ) = + b1 (xi ) + b2 (xi ), i = 0, . . . , n,
h2 h
6 3
4 (xi ) = b3 (xi ) + b4 (xi ) + b5 (xi ), i = 0, . . . , n,
h2 h
12 0
1 (xi ) = + a1 (xi ) + 4a2 (xi ), i = 0, . . . , n,
h2 h
12 0
2 (xi ) = a3 (xi ) + a4 (xi ) + 4a5 (xi ), i = 0 , . . . , n,
h2 h
12 0
3 (xi ) = + b1 (xi ) + 4b2 (xi ), i = 0 , . . . , n,
h2 h
12 0
4 (xi ) = b3 (xi ) + b4 (xi ) + 4b5 (xi ), i = 0, . . . , n,
h2 h
6 3
1 (xi ) = + a1 (xi ) + a2 (xi ), i = 0 , . . . , n,
h2 h
760 N. Caglar, H. Caglar / Computers and Mathematics with Applications 57 (2009) 757762

Fig. 1. Results for Example 1 with u(x) = x2 x.

Fig. 2. Results for Example 1 with v(x) = x x2 .

Table 2
The maximum absolute errors for Examples 1 and 2.
Example n Absolute error u(x) Absolute error v(x)

1 21 3.413935800722 1015 1.845745778439 1015


2 21 1.895787997540 103 9.605012467053 105
2 41 4.746227547210 104 2.408660889305 105
2 61 2.109990823906 104 1.071126800022 105

6 3
2 (xi ) = a3 (xi ) + a4 (xi ) + a5 (xi ), i = 0, . . . , n,
h2 h
6 3
3 (xi ) = + b1 (xi ) + b2 (xi ), i = 0, . . . , n,
h2 h
6 3
4 (xi ) = b3 (xi ) + b4 ( xi ) + b5 (xi ), i = 0 , . . . , n.
h2 h
To illustrate the method and error analysis of the boundary value problem (1), two examples are discussed in the
following section.

3. Computational results and discussion

In this section, some numerical examples are studied to demonstrate the accuracy of the present method. The results
obtained by the method are compared with the analytical solution of each example. It was found that our method in
comparison with the method of J. Lu [1] is much better with a view to accuracy and utilization.
N. Caglar, H. Caglar / Computers and Mathematics with Applications 57 (2009) 757762 761

Fig. 3. Results for Example 2 with u(x) = sin( x).

Fig. 4. Results for Example 2 with v(x) = x2 x.

Table 3
Comparison of our methods with the method of J. Lu in the solution of problem 2 for u(x).
1
x Exact solution J. Lu solution Proposed method h = 41

0.0 0.00000 0.00000 0.00000


0.1 0.30902 0.30935 0.30888
0.2 0.58779 0.59030 0.58751
0.3 0.80902 0.81686 0.80863
0.4 0.95106 0.96764 0.95060
0.5 1.00000 1.02770 0.99952
0.6 0.95106 0.98974 0.95060
0.7 0.80902 0.85491 0.80863
0.8 0.58779 0.63265 0.58751
0.9 0.30902 0.33996 0.30887
1.0 0.00000 0.00000 0.00000

Example 1. Consider the following equations

u00 (x) + xu(x) + xv(x) = f1 (x),


v 00 (x) + 2xv(x) + 2xu(x) = f2 (x),

subject to the boundary conditions u(0) = u(1) = 0, v(0) = v(1) = 0 where 0 < x < 1, f1 (x) = 2 and f2 (x) = 2. The
exact solutions u(x), v(x) are x2 x and x x2 , respectively. The observed maximum absolute errors for various values of
n are given in Table 2. The numerical results are illustrated in Figs. 1 and 2.
762 N. Caglar, H. Caglar / Computers and Mathematics with Applications 57 (2009) 757762

Example 2. We consider the following equations


u00 (x) + (2x 1)u0 (x) + cos( x)v 0 (x) = f1 (x),
v 00 (x) + xu(x) = f2 (x),
subject to the boundary conditions u(0) = u(1) = 0, v(0) = v(1) = 0 where 0 < x < 1, f1 (x) = 2 sin( x) + (2x
1) cos( x) + (2x 1) cos( x) and f2 (x) = 2 + x sin( x).The exact solutions u(x), v(x) are sin( x) and x2 x, respectively.
The observed maximum absolute errors for various values of n are given in Table 2. The numerical results are illustrated in
1
Figs. 3 and 4. The comparison of the B-spline solution of this example for h = 41 with the method of J. Lu [1] is given in
Table 3. As is evident from the numerical results, the present method approximates the exact solution very well.

References

[1] J. Lu, Variational iteration method for solving a nonlinear system of second-order boundary value problems, Comput. Math. Appl. 54 (2007) 11331138.
[2] S.H. Chen, J. Hu, L. Chen, C.P. Wang, Existence results for n-point boundary value problem of second order ordinary differential equations, J. Comput.
Appl. Math. 180 (2005) 425432.
[3] X.Y. Cheng, C.K. Zhong, Existence of positive solutions for a second order ordinary differential system, J. Math. Anal. Appl. 312 (2005) 1423.
[4] A. Lomtatidze, L. Malaguti, On a two-point boundary value problem for the second order ordinary differential equations with singularities, Nonlinear
Anal. 52 (2003) 15531567.
[5] H.B. Thompson, C. Tisdell, Boundary value problems for systems of difference equations associated with systems of second-order ordinary differential
equations, Appl. Math. Lett. 15 (6) (2002) 761766.
[6] H.B. Thompson, C. Tisdell, The nonexistence of spurious solutions to discrete, two-point boundary value problems, Appl. Math. Lett. 16 (1) (2003) 7984.
[7] F.Z. Geng, M.G. Cui, Solving a nonlinear system of second order boundary value problems, J. Math. Anal. Appl. 327 (2007) 11671181.
[8] N. Caglar, H. Caglar, B-spline solution of singular boundary value problems, Appl. Math. Comput. 182 (2006) 15091513.